LiquidAI Launches 3B Vision-Language Model LFM2.5-VL, Outscoring Larger Rivals

JosephJacks_ · x · 2026-08-13

LiquidAI has released LFM2.5-VL-3B, a lightweight vision-language model designed for fast instruction following, excellent grounding, and OCR capabilities. Built on the LFM2.5-2.6B base with a SigLIP2 400M vision encoder, it was pre-trained on approximately 34T tokens.

Despite its compact 3B parameter size, the model achieves comparable or better scores than models up to 2.6x its size:

The model is now available on Hugging Face, alongside a WebGPU demo for in-browser testing.

Related event: LiquidAI Releases LFM2.5-VL-3B Edge Multimodal Model(7 posts)→

Original post →

More from Models

Models channel →